Press a key. Speak. Done. VoiceBar lives in your menu bar — invisible until you need it. Powered by Whisper AI, running entirely on your Neural Engine. Nothing leaves your Mac. Ever.
macOS 14 Sonoma or later · Apple Silicon recommended · ~650 MB model download
You open a speech-to-text app. It asks you to pick a model. It offers twelve modes. It has a history panel, an AI rewrite button, a theme picker, a custom vocabulary editor. You just wanted to send an email.
I tried them all — WisprFlow, SuperWhisper, Spokenly, Pindrop, FluidVoice, AquaVoice, Handy, VoicePrism. They are great apps. But not for someone who wants to speak and have words appear, with zero decisions in between.
"All I want is to talk to my Mac, now, and get the result. I don't want to play with settings, decide on a model, or study the features. Just speak. Done."
VoiceBar does exactly one thing. No modes. No history. No AI rewrites. No themes. Press your hotkey, speak, your words land where your cursor is. Then it gets out of the way.
If it takes longer than 10 seconds to explain, it's too complicated. This one takes about 5.
That first icon on the left — VoiceBar. It lives in your Applications folder like any app, but hides from the Dock by default. The menu bar is all you need.
VoiceBar lives in your Applications folder like any app — but by default it hides from the Dock and runs from the menu bar. Press your hotkey and it wakes up instantly. If you ever need to tweak settings, open it from the menu bar or directly from Applications.
Talk at whatever speed you normally speak. Press Escape to cancel. Press the hotkey again to stop and transcribe.
Your text is inserted directly into whichever app you had open — email, Slack, Notes, browser, anything. No clipboard, no switching.
We say this with love. Siri is great. But she gets tired — especially if you're not speaking King's (or Queen's) English, or you're just in a flow and need it to not fail.
VoiceBar uses Whisper Large Turbo — the same model behind most professional transcription services — running entirely on your Neural Engine. Your Mac does all the work. In private.
Text lands directly in the focused field — not the clipboard, not a separate panel. Right there, where you were already typing.
Email, Slack, Notes, browser, code editor. If there's a text field, VoiceBar types into it.
Whisper downloads once (~650 MB) and runs on your Neural Engine. No internet needed after setup.
Changed your mind mid-sentence? Press Escape. Nothing is inserted, nothing is saved. Clean slate.
Finish speaking, then click into any window within 3 seconds. VoiceBar pastes automatically.
Installs like any Mac app and lives in your Applications folder. By default it hides from the Dock and tucks into the menu bar — your screen stays clean. Open it anytime from the menu bar icon or from Applications when you need to.
Default is ⌥Space. Change it to anything you like. Set it once, forget settings exist entirely.
An app that does one simple thing — accurately — should be free. Your Mac already has the hardware to run it. There's no server cost. There's nothing to charge for.
We find it hard to justify a subscription for something this basic, even if it feels like magic the first time you use it. Dictation should just be part of how you use your computer.
Not to our servers. Not to anyone's. There are no servers. VoiceBar doesn't have a backend — because it doesn't need one.
Download and use. No sign-up, no email, no profile, no onboarding wizard. Just open it.
We don't receive your audio, your transcriptions, or your usage data. Not anonymised. Not aggregated. Nothing.
The Whisper model downloads from Hugging Face once. That is the only time VoiceBar touches the internet.
Don't take our word for it. Every line of code is on GitHub. Read it, build it yourself, or just check.
Download. Grant mic access. Speak. If it's not for you, drag it to the bin. No accounts, no cancellations, no nothing.
Download for Mac — FreemacOS 14 Sonoma or later · Open source on GitHub